A Spatial-Temporal Comparison of EV Charging Station Clusters Leveraging Multiple Validity Indices

René Richard, Hung Cao, Monica Wachowicz

Abstract: Decoupling vehicles from the immediate consumption of fossil fuels introduces new opportunities in supporting sustainable mobility. Fostering a shift from vehicles with internal combustion engines to Electric Vehicles (EV) often involves using publicly funded subsidies. Given early EV adoption challenges, some charging stations may be under-utilized, others will serve a disproportionate number of users. An understanding of EV charging patterns is crucial for optimizing charging infrastructure placement and managing costs. Clustering has been used in the energy domain to ensure service continuity and consistency. However, clustering presents challenges in terms of algorithm and hyperparameter selection in addition to pattern discovery validation. The lack of ground truth information, which could objectively validate results, is not present in clustering problems. Therefore, it is difficult to judge the effectiveness of different modelling decisions since there is no external validity measure available for comparison. This work proposes a clustering process that allows for the creation of relative rankings of similar clustering results that will assist practitioners in the smart grid sector. The approach supports practitioners by allowing them to compare a clustering result of interest against other similar groupings over multiple temporal granularities. The efficacy of this analytical process is demonstrated with a case study using real-world EV charging event data from charging station operators in Atlantic Canada.